Support Questions
Find answers, ask questions, and share your expertise

Yarn AM used core keeps going up

New Contributor

Yarn Version : 3.1.1 

MR : 3.1.1

HDFS : 3.1.1

Timeline service : v1.5


i am using yarn capacity scheduler and 

facing workflow pending issue repeatedly

workflow pended because there is no resource for AM

but there are enough resources in the cluster for the AM

I find Ambari metrics -> Yarn -> AM used resource keeps going up

and reaches AM Vcore limit but there are no running workflow in cluster 


i think there are two possibilities

1. workflow finished but AM is not terminated 

2. workflow & AM finished but timeline service can't get finish information


i am restarting yarn over and again to reset AM used vcore

How can i fix this problem

Please help me

; ;